This master thesis project has been developed in the Italian headquarter of the bank Crédit Agricole, based in Parma. It designs a workflow to identify clients in different life moments, with the goal of improving the advertising strategies of the bank. In particular, it focuses on the marriage. A raw list of married people is obtained scraping the incoming bank transfers and is then refined using exploratory data analysis and network science tools; a wedding date is estimated for each of them and a list of not married people is also extracted. Afterwards, the dataset is created through feature engineering, using manual extractions and Bag-of-Words techniques. A Random Forest is exploited for feature selection and different models are eventually trained, tested and compared for classification.
This master thesis project has been developed in the Italian headquarter of the bank Crédit Agricole, based in Parma. It designs a workflow to identify clients in different life moments, with the goal of improving the advertising strategies of the bank. In particular, it focuses on the marriage. A raw list of married people is obtained scraping the incoming bank transfers and is then refined using exploratory data analysis and network science tools; a wedding date is estimated for each of them and a list of not married people is also extracted. Afterwards, the dataset is created through feature engineering, using manual extractions and Bag-of-Words techniques. A Random Forest is exploited for feature selection and different models are eventually trained, tested and compared for classification.
MACHINE LEARNING BASED APPROACH FOR BANKING CLIENTS’ LIFE MOMENTS DETECTION
CRUDELE, MICHELE MARIA
2021/2022
Abstract
This master thesis project has been developed in the Italian headquarter of the bank Crédit Agricole, based in Parma. It designs a workflow to identify clients in different life moments, with the goal of improving the advertising strategies of the bank. In particular, it focuses on the marriage. A raw list of married people is obtained scraping the incoming bank transfers and is then refined using exploratory data analysis and network science tools; a wedding date is estimated for each of them and a list of not married people is also extracted. Afterwards, the dataset is created through feature engineering, using manual extractions and Bag-of-Words techniques. A Random Forest is exploited for feature selection and different models are eventually trained, tested and compared for classification.File | Dimensione | Formato | |
---|---|---|---|
Michele_Crudele_MSc_Thesis.pdf
accesso riservato
Dimensione
8.19 MB
Formato
Adobe PDF
|
8.19 MB | Adobe PDF |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/36021